Cohere fights copyright lawsuit

Toronto-based AI firm Cohere is seeking dismissal of a copyright infringement lawsuit filed by several media publishers in a U.S. court. Cohere argues that the plaintiffs, including the Toronto Star, deliberately misused its AI tools to fabricate evidence. The company claims the publishers crafted specific prompts to elicit copyrighted material, including inaccuracies, and that no real-world user has infringed copyright using Cohere's software.

The lawsuit, initiated in February, accuses Cohere of repurposing articles and generating false information under the publishers' names, thereby undermining their brands and profits. The media outlets are seeking an injunction to prevent Cohere from using their content to train AI models and demanding $150,000 for each allegedly scraped article. Cohere defends its practices, asserting its commitment to responsible AI development and respect for intellectual property laws, stating that the publishers misused a developer demo tool to engineer the alleged infringements.

Cohere's motion highlights that its AI solutions are designed to analyse internal data for businesses, not to replace news consumption. The company emphasises that the media outlets' claims are based on manufactured data and do not reflect actual usage of its enterprise-focused AI tools.